Case study · AI Engineering
A Swiss insurance group
Claims arrived as PDFs, photos and emails — then got retyped by hand. We automated intake with AI, with human review only where it matters.
The mess we walked into
Claims intake was a manual funnel: documents arrived in every imaginable format, clerks retyped the data, and errors surfaced weeks later during settlement. Peak periods meant overtime and backlogs; compliance required every step to be traceable.
What we did
- AI-powered intake: classification and data extraction across document types
- Human-in-the-loop review queue for low-confidence and high-value claims
- Full audit trail for every automated decision — compliance-first design
- Integrated into the existing claims system; no workflow retraining
Before / after
Intake per claim manual retyping automated, reviewed by exception
Peak periods overtime and backlog same team, steady flow
Errors found at settlement caught at intake
Traceability manual notes automatic audit trail
← All case studies"The backlog after storm season used to take weeks. Now the system handles the volume and our people handle the judgment calls."
Head of Claims Operations, insurance group, Switzerland
A similar system? A similar mess?
Tell us what you're dealing with. We back every claim on this page up on a due-diligence call.